<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<div class="moz-cite-prefix">Hi Pradeep,<br>
<br>
Glad you are up and running! Let me see if I can answer your
remaining questions...
<blockquote type="cite">
<ol>
<li>What did you mean by "Production setting"?<br>
</li>
</ol>
</blockquote>
I just wanted you to be careful since I used "xhost +" to give
pvserver access to the GPU. That command weakens/compromises x11
security by allowing *anyone* to use the local x server. It's the
easiest way to get PV to use GPU on a system that already has x11
running. I just want to make it easy for you to get something
going. Once you have it working you can refine the x11 options so
that security is not compromised.<br>
<br>
<blockquote type="cite">
<ol>
<li>I am right now testing PV on small test cases over
machines (my Mac and the remote Linux) which have both
graphics hardware. For such a case why do I need "reverse
connection"? Should I always use "reverse connection"?<br>
</li>
</ol>
</blockquote>
The PV connection type doesn't have anything to do with if you are
using graphics hardware or not, it simply describes who connects
to who when. ParaView's reverse connection is its most versatile
connection option. It can handle every use case I've ever come
across. Here's the difference between forward and reverse: In the
forward connection the client attempts and fails immediately if
the server isn't up and ready for it. however with the reverse
connection the client waits for the server to start and connect
back. The case where you really need the reverse connection is
when a batch system is involved and the server doesn't start up
immediately. <br>
<br>
Using graphics hardware correctly on the server is another issue.
For a normal linux box with x11 running you just need to tell the
xserver that it's ok to let pv to use the graphics cards. for a
cluster you may have to start the xserver yourself. and remember
don't use ssh x forwarding with pv!<br>
<br>
<blockquote type="cite">
<ol>
<li>When I am at lab, I have a static IP for my Mac and I can
log into it from other computers. But I cannot access my Mac
from outside when I am not in the lab. Does this mean I
cannot use pvserver outside the lab? Or is <a
moz-do-not-send="true"
href="https://hpcforge.org/plugins/mediawiki/wiki/pv-meshless/index.php/Launching_ParaView_on_HPC_Machines#Step_2"
target="_blank">this link</a> is talking about this
problem?</li>
</ol>
</blockquote>
What you need is some path through the network to the machine
where the pvserver will run. As long as you can ssh to a machine
that can see the pvserver machine you will be able to make it work
using ssh tunnels and/or port forwarding. ssh is extremely
versatile. I don't have enough info about your network to give you
a more specific answer.<br>
<br>
<blockquote type="cite">
<ol>
<li>
<div>Once I manage to understand and get it working, I want
to use PV using a supercomputing facility. On the
supercomputing facility I can install PV on my login node
(which has 64 processors) but not on the supercomputing
cluster. <span
style="font-family:arial,sans-serif;font-size:13px">If I
want to use the cluster (which has around 50,000
processors) to visualize a big </span></div>
<div><span
style="font-family:arial,sans-serif;font-size:13px">data
(around 800GB if possible), will installing Paraview
with MPI support on the login node, which is accesible
by the cluster, be sufficient? </span></div>
<div>
</div>
</li>
</ol>
</blockquote>
It's not as bad as you think ;-) Use the center provided MPI
libraries and build or install PV in a folder on a filesystem that
is mounted on the compute nodes. Often your home folder is mounted
on compute nodes. If not, there's usually a scratch file system
for parallel I/O mounted. You'll need to launch the server through
a batch script. For debugging purposes you could use an
interactive batch job to get the hang of it.<br>
<br>
Hope this clarifies<br>
Burlen<br>
<br>
<br>
On 01/30/2013 03:41 AM, Pradeep Jha wrote:<br>
</div>
<blockquote
cite="mid:CAL_eiiTziko6H+XPBzsqF5tjh2-hma5wfSirKHhX8SgMdiV-yA@mail.gmail.com"
type="cite">
<div dir="ltr">Hello Burlen,
<div><br>
</div>
<div>I managed to get Paraview talk to the server using the
instructions on <a moz-do-not-send="true"
href="https://hpcforge.org/plugins/mediawiki/wiki/pv-meshless/index.php/Launching_ParaView_on_HPC_Machines#Step_2">this
website</a>. That itself clarified Question number 1, 4 and
5 for me. I would really appreciate if you can still answer my
other queries.</div>
<div><br>
</div>
<div style="">Thanks again,</div>
<div style="">Pradeep</div>
<div><br>
</div>
<div style=""><br>
</div>
</div>
<div class="gmail_extra"><br>
<br>
<div class="gmail_quote">2013/1/30 Pradeep Jha <span dir="ltr"><<a
moz-do-not-send="true"
href="mailto:pradeep@ccs.engg.nagoya-u.ac.jp"
target="_blank">pradeep@ccs.engg.nagoya-u.ac.jp</a>></span><br>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">
<div dir="ltr">Hello Burlen,
<div><br>
</div>
<div>thanks for the detailed response. I am still not able
to get it work and somethings are still not clear to me.
I want to ask some very fundamental questions as this is
my first time trying to set up a something over the
networks myself and the online instructions are a bit
too technical for me. </div>
<div><br>
</div>
<div>Presently, I have PV 3.98 installed from source with
MPI support on my local Mac and a remote Linux machine.
I ran the pvsc file that you sent from my local Mac but
I was not sure what goes in the input for "MPI Root" and
"ParaView Root". I left those two fields blank and tried
to connect and gave me the following error:</div>
<div><br>
</div>
<div>------------------------------------------------------------------------</div>
<div>
<div>pradeep@laptop subset]$MPI_NP=4</div>
<div>MPI_ROOT=</div>
<div>PV_ROOT=</div>
<div>PV_SERVER_PORT=11111</div>
<div>REMOTE_USER=pradeep</div>
<div>SERVER_HOST=<a moz-do-not-send="true"
href="http://83.shin.fluid.cse.nagoya-u.ac.jp"
target="_blank">83.shin.fluid.cse.nagoya-u.ac.jp</a></div>
<div>SSH_PATH=/usr/bin/ssh</div>
<div>XTERM_PATH=/usr/bin/xterm</div>
<div>Accepting connection(s): laptop.local:11111</div>
<div>Server launch command is : /usr/bin/xterm -T
"ParaView Server <a moz-do-not-send="true"
href="http://pradeep@83.shin.fluid.cse.nagoya-u.ac.jp:11111"
target="_blank">pradeep@83.shin.fluid.cse.nagoya-u.ac.jp:11111</a>"
-e /usr/bin/ssh -t -R 11111:localhost:11111 <a
moz-do-not-send="true"
href="mailto:pradeep@83.shin.fluid.cse.nagoya-u.ac.jp"
target="_blank">pradeep@83.shin.fluid.cse.nagoya-u.ac.jp</a>
DISPLAY=:0 xhost + ;
LD_LIBRARY_PATH=/lib:/lib/paraview-3.98/:LD_LIBRARY_PATH
DISPLAY=:0 /bin/mpirun -np 4 /bin/pvserver
--reverse-connection --server-port=11111
--client-host=localhost</div>
<div>The process failed to start. Either the invoked
program is missing, or you may have insufficient
permissions to invoke the program. </div>
<div>Server launch timed out. </div>
<div>---------------------------------------------------------------------------<br>
</div>
<div><br>
</div>
<div>Here are my questions:</div>
<div>
<ol>
<li>How do I get the above thing working?</li>
<li>What did you mean by "Production setting"?<br>
</li>
<li>I am right now testing PV on small test cases
over machines (my Mac and the remote Linux) which
have both graphics hardware. For such a case why
do I need "reverse connection"? Should I always
use "reverse connection"?<br>
</li>
<li>I don't understand when I initiate a pvserver on
the remote Linux machine and try to connect it
from local Mac, why doesn't it ask for a password?<br>
</li>
<li>My understanding is that when I get this
connection going, all my data should be on the
server. I simply use the local Mac GUI and I
should be able to visualize and browse through
data at the remote end. Is that correct?<br>
</li>
<li>When I am at lab, I have a static IP for my Mac
and I can log into it from other computers. But I
cannot access my Mac from outside when I am not in
the lab. Does this mean I cannot use pvserver
outside the lab? Or is <a moz-do-not-send="true"
href="https://hpcforge.org/plugins/mediawiki/wiki/pv-meshless/index.php/Launching_ParaView_on_HPC_Machines#Step_2"
target="_blank">this link</a> is talking about
this problem?</li>
<li>
<div>Once I manage to understand and get it
working, I want to use PV using a supercomputing
facility. On the supercomputing facility I can
install PV on my login node (which has 64
processors) but not on the supercomputing
cluster. <span
style="font-family:arial,sans-serif;font-size:13px">If
I want to use the cluster (which has around
50,000 processors) to visualize a big </span></div>
<div><span
style="font-family:arial,sans-serif;font-size:13px">data
(around 800GB if possible), will installing
Paraview with MPI support on the login node,
which is accesible by the cluster, be
sufficient? </span></div>
<div>
<br>
</div>
</li>
</ol>
<div>I guess these questions are extremely basic but I
am responsible for figuring this whole thing out
myself and with not much direct experience in
networking. So it is troubling me a bit.</div>
<div>
<br>
</div>
<div>Hoping to hear from you,</div>
<div>Pradeep</div>
</div>
</div>
</div>
<div class="HOEnZb">
<div class="h5">
<div class="gmail_extra"><br>
<br>
<div class="gmail_quote">2013/1/30 Burlen Loring <span
dir="ltr"><<a moz-do-not-send="true"
href="mailto:bloring@lbl.gov" target="_blank">bloring@lbl.gov</a>></span><br>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">Hi
Pradeep,
<div><br>
<br>
<blockquote class="gmail_quote" style="margin:0
0 0 .8ex;border-left:1px #ccc
solid;padding-left:1ex">
vtkClientSocket (0x7fc9ae108cb0): Socket error
in call to connect. Permission denied.<br>
</blockquote>
<br>
</div>
Your connection is being blocked somewhere in
between your compute node and workstation. there
are various configuration setting on either/both
client and server that could cause it. Fortunately
we do not need to change any of these settings ,
many of which require root access and potentially
open security vulnerabilities. Instead, you will
use an ssh tunnel and a server config (pvsc)
tailored to your situation to automate the
process.<br>
<br>
I'm attaching a minimal pvsc that illustrates how
one might configure a reverse connection to a
server with graphics hardware. This pvsc is for
illustration only, don't use in a production
setting! There are number of liberties I've taken,
for example I assume that X11 is already running,
and I use "xhost +"(very very bad) to enable
pvserver to access gpu's. Normally I would put all
of the server side stuff in a shell script. I
didn't do that here to keep things simple for you.
I hope you can use this to understand how PV
works.
<div>
<br>
<br>
<blockquote class="gmail_quote" style="margin:0
0 0 .8ex;border-left:1px #ccc
solid;padding-left:1ex">
1) I connect to the CentOS machine using "ssh
-X machinename".<br>
</blockquote>
</div>
Don't do that. With -X forwarding you won't be
taking advantage of your cluster's graphics
hardware.<br>
<br>
Hope this helps<span><font color="#888888"><br>
Burlen</font></span>
<div>
<div><br>
<br>
On 01/29/2013 06:41 AM, Utkarsh Ayachit wrote:<br>
<blockquote class="gmail_quote"
style="margin:0 0 0 .8ex;border-left:1px
#ccc solid;padding-left:1ex">
Doesn't look like your client machine can
connect to the server<br>
machine. From the client machine try:<br>
<br>
<blockquote class="gmail_quote"
style="margin:0 0 0 .8ex;border-left:1px
#ccc solid;padding-left:1ex">
ping <a moz-do-not-send="true"
href="http://83.shin.fluid.cse.nagoya-u.ac.jp"
target="_blank">83.shin.fluid.cse.nagoya-u.ac.jp</a><br>
telnet <a moz-do-not-send="true"
href="http://83.shin.fluid.cse.nagoya-u.ac.jp"
target="_blank">83.shin.fluid.cse.nagoya-u.ac.jp</a>
11111<br>
</blockquote>
<br>
<br>
On Tue, Jan 29, 2013 at 4:58 AM, Pradeep Jha<br>
<<a moz-do-not-send="true"
href="mailto:pradeep@ccs.engg.nagoya-u.ac.jp"
target="_blank">pradeep@ccs.engg.nagoya-u.ac.jp</a>>
wrote:<br>
<blockquote class="gmail_quote"
style="margin:0 0 0 .8ex;border-left:1px
#ccc solid;padding-left:1ex">
Hello,<br>
<br>
I have installed Paraview 3.98 on my
Desktop (Mac OSX 10.8) and a remote<br>
machine running CentOS 5.4. The CentOS
machine has rendering hardware so I<br>
havent installed PV with OSMesa support.<br>
<br>
I can login to the CentOS from my Mac
using SSH and vice versa.<br>
<br>
Here is how I am trying to connect:<br>
<br>
1) I connect to the CentOS machine using
"ssh -X machinename". When I start<br>
the "pvserver" on the CentOS the output
is:<br>
-------------------------------------------------------------------<br>
[pradeep@83 ~]$pvserver<br>
Waiting for client...<br>
Connection URL: cs://<a
moz-do-not-send="true"
href="http://83.shin.fluid.cse.nagoya-u.ac.jp:11111"
target="_blank">83.shin.fluid.cse.nagoya-u.ac.jp:11111</a><br>
Accepting connection(s): <a
moz-do-not-send="true"
href="http://83.shin.fluid.cse.nagoya-u.ac.jp:11111"
target="_blank">83.shin.fluid.cse.nagoya-u.ac.jp:11111</a><br>
-------------------------------------------------------------------<br>
<br>
2) Then I add this information in the
paraview I am running on my Mac as<br>
shown in the attached image file.<br>
<br>
3) I click on connect.<br>
<br>
I get the following error:<br>
------------------------------------------------------------------------<br>
ERROR: In /Users/pradeep/softwares/ParaView/VTK/Common/System/vtkSocket.cxx,<br>
line 481<br>
vtkClientSocket (0x7fc9ae108cb0): Socket
error in call to connect.<br>
Permission denied.<br>
-----------------------------------------------------------------------<br>
<br>
I dont have any expertise in networking
and so I dont know much about<br>
firewalls of my systems here nor do any my
colleagues do.<br>
<br>
Thanks in advance,<br>
Pradeep<br>
<br>
<br>
<br>
_______________________________________________<br>
Powered by <a moz-do-not-send="true"
href="http://www.kitware.com"
target="_blank">www.kitware.com</a><br>
<br>
Visit other Kitware open-source projects
at<br>
<a moz-do-not-send="true"
href="http://www.kitware.com/opensource/opensource.html"
target="_blank">http://www.kitware.com/opensource/opensource.html</a><br>
<br>
Please keep messages on-topic and check
the ParaView Wiki at:<br>
<a moz-do-not-send="true"
href="http://paraview.org/Wiki/ParaView"
target="_blank">http://paraview.org/Wiki/ParaView</a><br>
<br>
Follow this link to subscribe/unsubscribe:<br>
<a moz-do-not-send="true"
href="http://www.paraview.org/mailman/listinfo/paraview"
target="_blank">http://www.paraview.org/mailman/listinfo/paraview</a><br>
<br>
</blockquote>
_______________________________________________<br>
Powered by <a moz-do-not-send="true"
href="http://www.kitware.com"
target="_blank">www.kitware.com</a><br>
<br>
Visit other Kitware open-source projects at
<a moz-do-not-send="true"
href="http://www.kitware.com/opensource/opensource.html"
target="_blank">http://www.kitware.com/opensource/opensource.html</a><br>
<br>
Please keep messages on-topic and check the
ParaView Wiki at: <a moz-do-not-send="true"
href="http://paraview.org/Wiki/ParaView"
target="_blank">http://paraview.org/Wiki/ParaView</a><br>
<br>
Follow this link to subscribe/unsubscribe:<br>
<a moz-do-not-send="true"
href="http://www.paraview.org/mailman/listinfo/paraview"
target="_blank">http://www.paraview.org/mailman/listinfo/paraview</a><br>
</blockquote>
<br>
</div>
</div>
</blockquote>
</div>
<br>
</div>
</div>
</div>
</blockquote>
</div>
<br>
</div>
</blockquote>
<br>
</body>
</html>